Join our daily and weekly newsletters for the latest updates and the exclusive content on AI coverage. Learn more
The graphics processing units (GPU), the costly computer flea made by Nvidia, AMD and Sima.a. Are no longer the only way to train and deploy artificial intelligence.
Organic black box (BBB)A startup founded in Baltimore developing a new class of AI equipment, is emerged from stealth with its Bionode platform – a computer system which integrates living and cultivated neurons in laboratory with traditional processors.
The company, which has worked quietly while filing patents and refining its technology, believes that its organic computer approach – the growth of new neurons specifically to act as computer chips using donor human stem cells and rats -derived cells – could offer a low -power adaptive alternative to conventional GPUs.
“In the past 20 years, three independent areas – biology, equipment and calculation tools – have advanced to the point where organic computer science is now possible,” said Alex Ksendzovsky, co -founder and CEO of BBB, in a video interview with Venturebeat.
Member of the NVIDIA creation incubator, BBB is positioned as an increase and an increase in AI chips based on dominant silicon that Nvidia and others produce.
By taking advantage of the capacity of neurons to reclaim physically, the company aims to reduce energy costs, improve the efficiency of treatment and accelerate the formation of the AI model – the layers which have become increasingly urgent as the adoption of AI develops.
It is not science fiction, despite the incredible premise: BBB neural chips already feed computer vision and LLM for customers, and have entered conferences with two partners to concede to its technology for computer vision applications – although the company has refused to specifically appoint its customers and partners, citing confidentiality agreements. He also accepts surveys of potential partners and customers website.
Mix biology and equipment
At the heart of the BBB approach is the Bionode platform, which uses cultivated neurons in the laboratory wired in computer systems.
“We have several models that we use,” said Ksendzovsky. “One of these models comes from rat cells. One of these models comes from truly human stem cells which are converted into neurons. ”
The co-founder said that “hundreds of thousands of them” are integrated into a dish containing 4,096 electrodes, which is the basis of a Bionode chip. He also said they lived for more than a year before having to be replaced.
The idea is to exploit the natural adaptability of neurons to the processing of AI, creating a hybrid computer system which differs fundamentally from rigid fleas and based on today’s transistors.
Ksendzovsky, who has worked with neurons on electrodes since 2005, initially entertained the idea of using them to predict the stock market. His mentor, Dr Steve Potterrejected the idea at the time.
“Why don’t we use neurons to predict the stock market so that we can all be rich?” Ksendzovsky remembers asking Potter, who laughed as impractical. “At the time, he was right,” admitted Ksendzovsky.
Since then, improvements in electrodes technology, calculation tools and the longevity of neurons have made the biological calculation viable. “The biological network has evolved more than hundreds of millions of years in the most effective computer system ever created,” said Ksendzovsky.
This configuration offers two immediate advantages:
• More efficient computer vision: Bionode has been tested as a layer of pre -treatment for AI classification tasks, reducing both inference time and GPU energy consumption.
• Accelerated training of the large language model (LLM): Unlike GPUs, which require frequent recycling cycles, neurons adapt to the fly. This could considerably reduce the time and energy necessary for updating LLMS, by attacking a key bottleneck in the AI scale.
“One of our greatest breakthroughs is to use biological networks more effectively to form important language models (LLM), which reduces massive energy consumption required today,” said Ksendzovsky.
Build a viable and living GPU with the help of Nvidia
The NVIDIA GPUs played a decisive role in the rapid AI progress, but their high energy consumption and their growing cost aroused concerns about scalability.
BBB sees the opportunity to introduce an alternative more economical in power while operating in the Nvidia ecosystem.
“We do not consider ourselves as direct competitors to Nvidia, at least in the near future,” noted Ksendzovsky. “Organic IT and silicon IT will coexist. We still need GPU and processors to process data from neurons. ”
In fact, according to the co-founder, “we can use our biological networks to increase and improve silicon-based AI models, which makes them more precise and more energy efficient.”
The long-term vision of AI material, he argued, will be a modular ecosystem, where organic computers, silicon fleas and even quantum computer science each play a role.
“The future of computer science will be a modular ecosystem where traditional silicon, organic IT and quantum computer science each play a role based on their strengths,” he said.
Although BBB has not yet disclosed a commercial launch date, the Baltimore, Maryland, company Maryland in the Bay region when it is preparing to evolve its technology.
The future of hybrid AI treatment
While silicon-based GPU remains the industry standard, the Brain-on-A-Chip concept of BBB gives an overview of a future where IA equipment is no longer limited to transistors and circuits.
The capacity of neurons to be reconfigured dynamically could allow AI systems which are more energy efficient, adaptive and capable of continuous learning.
“We already apply a biological calculation to computer vision. We can code images in a biological network, let the neurons treat them, then decode the neuronal response to improve the precision of the classification,” said Ksendzovsky.
Beyond efficiency gains, BBB also believes that its biological approach can provide more in-depth information on how AI models process data.
“We have built a closed loop system that allows neurons to reclassify, increasing the efficiency and precision of AI tasks,” he said.
Despite the potential, Ksendzovsky acknowledges that ethical considerations will be a discussion in progress. BBB is already working with ethicians and regulatory experts to ensure that its technology is developed in a responsible manner.
“We don’t need millions of neurons to treat the whole environment as a brain does. We only use what is necessary for specific tasks, keeping in mind ethical considerations, “he said.
BBB bets that living tissues, not just silicon, could be the key to the next Jump of AI forward.
